Best practice in surveys
Guidance for students and staff carrying out surveys. Find advice about survey design, promotion and follow up in order to achieve a high response rate.
Creating a successful survey
Surveys are successful when:
- using a survey is the best method of capturing data
- the questions are well-designed
- the survey has a high-response rate
- the results are communicated to an appropriate audience
- participants are made aware of the results and any corresponding actions that will follow.
Where applicable, please also ensure that you comply with the University's ethical principles for research.
How to design an effective online survey
Prepare your survey in advance
Do a complete outline of your survey in a text editor before even logging in to an online survey tool, such as the JISC Online Survey, to start building it. This will give you the opportunity to consider the wording of each question, the full set of responses, possible ambiguities in questions and the overall structure of a survey. Good structural design is important in any survey. Make sure the questionnaire is interesting and topical and one that participants will want to complete. Ensure you pilot the survey in advance.
Keep survey size to a minimum
The simple fact is the bigger your survey is the less likely respondents will be to complete it. Does the questionnaire already exist, or must you create one? Check the University’s survey wiki to see if previous research will give you the answers you require. If you must create another survey, then pare down your survey so that you ask the crucial questions that you want answers to, and get rid of questions that are really supplemental to your current research aims. You can always add a question at the end, asking respondents if they would mind taking part in further surveys, and send them out supplemental surveys at a later date. Streamlined surveys get the best results.
Structure your surveys into logical chunks
Give your survey a good structure. On the first page, make it clear what the aim of the survey is. Generally, keep personal identification questions to the end. Above all, break your survey up into logically ordered pages. This helps you group relevant questions together, makes the survey less daunting (one long page can be extremely off-putting to respondents), and also gives you flexibility in creating conditional pages, which will not show questions that are irrelevant to certain respondents.
Don't confuse multiple and single choice options
A common mistake in survey design is using a multiple choice question style for a question where you are only seeking a single answer. Think carefully for each question whether you want respondents to select just a single option, or whether you want to leave open the possibility of several responses. For example, consider the difference between these two questions:
- What age group do you fall in? 15-25, 26-45, 46-65, 66+ (single choice)
- What continents have you holidayed on? Europe, Americas, Africa, Asia, Oceania, None (multiple choice)
Use images sparingly
When using images in surveys, use them sparingly! The University logo at the top of a survey can give it authority, and multiple choice image questions, where appropriate, make the survey more interesting for respondents. If the Survey is supported by the Students’ Union, request to use their logo alongside that of the institution. Make sure you resize large graphics to smaller, more easily downloadable sizes. Check to see if your survey provider allows you to resize images that you've uploaded, rather than having to adjust them in software like Photoshop.
Be aware of copyright and data protection laws
You don't need to be an expert on copyright or data protection laws, but you must always bear in mind that anything you include in your survey and any data you gather should pay due respect to the law. Do not infringe copyright on images, nor include anything that is potentially libellous. If you are collecting personal data, be aware of the data protection act, and ask for any required permissions for any uses you may make of the data gathered. Ensure you state the appropriate assurances about confidentiality and anonymity in the survey. Have all ethical issues been explored? Does your survey need to be approved by the departmental or University Research Ethics Committee?
The University’s website has guidance on data protection for all members of University staff.
How to get your survey read by participants
Set a goal for your response rate
As part of a communication and marketing plan, a goal for the response rate should be set, with planned interventions and promotional activities set throughout the fieldwork stage of the survey. Don’t just rely on the initial email and the subsequent reminder emails to promote the survey – create an integrated campaign involving a range of tools. These tools could potentially include word of mouth (from professional support staff, academics, student academic reps and students themselves), print materials (e.g. posters, flyers, leaflets, bookmarks, pens, beer mats), web sites, and social network sites (such as Twitter and Facebook).
Think carefully about the wording of your email invitation
Your email to a potential respondent has to feel like a personal invitation. The subject line has to show benefit or value to the respondent because that may be all they read. For example:
- The International Office needs your help
- Spill some secrets and you could win an iPad
- How are we doing?
- Is our departmental web site doing its job? We need your feedback
The text of the email should be brief and direct. If you know the first name of the recipient and can customise the emails then start off with “Dear Jennifer”. If not then say, “Dear student”, or whatever your relationship is to the recipient. Do NOT say "Dear friend" as your email immediately sounds like some shady offer and may be marked as spam. "Dear Jennifer Smith" has the same effect. In the body of the email:
- Explain why you are contacting them
- If there is an incentive then mention it
- Include a call to action ("Click the link below to start the survey")
- Remember to say "Thank you"
- Put a real person's name and job title at the end.
Keep it brief. People rarely read unsolicited emails more than once or bookmark them for later attention. You want them to read the email and act on it immediately.
Top and tail your survey
The survey itself has to follow the same rules as the email invitations. Have an opening page which thanks the respondent for taking the survey and tells them its purpose and how long it will take. If there is an incentive show a picture of it with a brief description. Put a real person's name and job title at the end of the first page if appropriate. If you need to include a set of rules for a promotion, competition or prize draw then put them on the second page. But make this page conditional and have a tick box on the first page for people who want to see them. At the end of the survey, once again thank people for completing it, remind them they need to press one more button to complete the survey and, if appropriate, be a contender for the incentive.
Send the email invitations at times that match your respondent's email habits
Emails which arrive at an email address on a Monday morning may be buried amongst the weekend's spam and may be easily lost (or inadvertently deleted). So try sending them out at another time, such as an Wednesday afternoon or a Tuesday morning, for example.
HTML or plain text?
HTML emails are increasing in acceptance and can make it easier to brand your email invitation as coming from a trusted source. Don't put any vital information in an image unless it is repeated in the text. However, many email programs don't show images automatically. Make sure the content and purpose of the email is clear without any images. Send your HTML emails as "multi-part". This means that anyone reading it on a mobile device or email client which doesn't display HTML automatically will still see the text of your email and not a load of HTML code.
Promoting a survey on your web site
When promoting a survey via links on a web page include all the courtesy, enticement and calls to action that should be included in an email invitation. Don't display a survey automatically. A visitor who has clicked a link to a survey (because of your carefully worded invitation) is far more likely to answer it honestly and complete it than one who is shown it involuntarily.
How to improve your survey response rates
Create and maintain an environment where students are encouraged to provide open and honest feedback on all aspects of their student experience. Senior management support is essential.
Plan feedback opportunities
Avoid ‘survey fatigue’ by planning how and when you will communicate with students about surveys that will take place during the year and any resulting actions.
Be prepared to respond quickly
In your annual planning, be ready to receive and act on student feedback within a short timescale (relative to the student academic year).
Communicate when things have changed
Students will respond to surveys if their experience is that things will change as a result. But they won’t know what has improved unless you tell them. Communicate changes you have made historically the University has used University-wide leaflets and posters reporting "You Said, We Have". Survey results and responses should never be invisible to the students, even if they present difficulties for you or the University.
Know what you are asking
Make sure that the questions in your survey are accurately worded and will give you the information you need. Always pilot surveys with a few students. Ask them afterwards in a focus group what they thought each question meant. It’s also helpful to ask students what they thought you should have asked them about!
Build on response rates
As response rates rise, results become more reliable. Decisions based on this information will be more accurate and changes should more closely reflect student needs. This creates a ‘virtuous circle’ of feedback and action that should encourage student participation in surveys in the future.
Use incentives, if appropriate
An incentive (such as a book token, an e-voucher or even real cash) can improve the response rate of your survey. At the margin, an incentive may just encourage some people to take the survey that wouldn't otherwise. Make the incentive appropriate to your audience. It's unlikely that anyone will be deterred from taking your survey because of an incentive. Don't make the incentive too big or else your survey will be circulated widely to people with no relevance to your survey that are just after the prize. Mention the incentive, and show a picture of the incentive, at appropriate places on the survey.